The results from David’s Primo are in, with a couple improvements to the methodology.
One, instead of a phone camera, I used a USB microscope and an OTG viewer app on a tablet, which affords a better illumination and better view. A unlike the phone, it doesn’t time out and go back to the home screen, at which point touching the phone can disturb the measurement.
Also I have a new extremely-compliant mechanism that allows one dimensional movement only, and a push-rod to move it, which is much less janky than what I was doing before.
I tried for a mechanical means to amplify the movement but that was a bust, so I am measuring at 1:1 scale.
One of the things I learned is that the spacing between the lines is uneven:
The spacing in the SVG file in Inkscape is of course perfect, and I rotated the artwork at an angle, assuming that would prevent any aliasing effects and the resulting edges would be perfectly spaced, but I was wrong and the pitch came out uneven. This has a big effect on the Vernier readout for individual increments but smoothing over multiple increments reduces the effect.
The results:
The movement of the corner block and belt near corner were too small to measure. I applied only about 3 lbs (and -3 lbs) and the movement was less than 0.025 mm which would put it at less than 0.01 mm/lb. Maybe much less. I don’t know.
The movement of the truck was 26.5 lbs per mm, or about 0.038 mm per lb.
I also measured the Y gantry rail in the center near the core, and the deflection was 19.3 lbs per mm, or about 0.052 mm per lb, which is slightly more deflection than the trucks and would imply 0.014 mm per lb of horizontal deflection of the rail. This is 1" stainless steel.
I also measured the Z rail just above the core (about 75 mm above the Y rail) and the deflection was 0.035 mm per lb. Deflection of the Z rail just below the core (about 100 mm below the Y rail) was about 0.089 mm per lb. This is inclusive of the lateral rail deflection and the core tilt by itself would be -0.017 above the core and 0.037 below the core.
So to summarize:
- Negligible deflection from corners and belt attachments
- 0.038 mm/lb from truck movement (belt stretch)
- 0.014 mm/lb from gantry rail bending
- 0.037 mm/lb from core tilt
- 0.089 mm/lb total from above 3 sources
At first I was trying to hurry so as not to impose on David, but I eventually figured out that he was having a leisurely afternoon, and we had a nice chat. Thanks, David for being such a gracious host!